Numerical Control (NC) machining is the cutting edge of modern manufacturing technology. NC errors could break cutter edges, destroy work pieces and even damage the machine tool. In recent years, more and higher speed cutting is applied in the industry due to the advancement of machine tool technology and the demand of shorter production time. However, checking the NC codes for high speed machining is difficult due to the lack of information on material removal rate and the large size of NC blocks. In this paper, a novel high speed NC simulation method in an extended Z-map approach is presented, which offers higher simulation accuracy of the resulted geometry and with reasonable cutting load calculation. In conclusion the authors propose the pervasive manufacturing modeling and simulation for multi machining and layered manufacturing processes.

1. Quick NC Simulation & Verification for High Speed Machining
Liu Peiling
(Singapore Institute of Manufacturing Technology, Singapore)
Ma Yongsheng
(School of MAE, Nanyang Technological University, Singapore)
Song Bin
(Singapore Institute of Manufacturing Technology, Singapore)
Koh Seng Chee
(Singapore Precision Engineering and Tooling Association, Singapore)
Abstracts: Numerical Control (NC) machining is the cutting edge of modern manufacturing technology. NC errors
could break cutter edges, destroy work pieces and even damage the machine tool. In recent years, more and higher
speed cutting is applied in the industry due to the advancement of machine tool technology and the demand of
shorter production time. However, checking the NC codes for high speed machining is difficult due to the lack of
information on material removal rate and the large size of NC blocks. In this paper, a novel high speed NC
simulation method in an extended Z-map approach is presented, which offers higher simulation accuracy of the
resulted geometry and with reasonable cutting load calculation. In conclusion the authors propose the pervasive
manufacturing modeling and simulation for multi machining and layered manufacturing processes.
Keywords: NC verification; NC simulation; NC post-processing; Collaborative machining
1. INTRODUCTION
Comparing with the sharp decline of the
computing cost, worldwide material and machine
tool prices have up surged significantly. Saving
material and making efficient use of machines
through pervasive computing become important in
the midst of increasing raw material and energy
costs. One NC error could make the work piece a
scrap; and it could take days to reproduce another
work piece. Due to the work piece’s material and
processing costs, machining errors could erode the
profit considerably. In the age of small batch
production, there is no time for “trials and errors”
approach. Therefore, verifying and optimizing
precision NC machining have significant impacts on
the profitability of a company. The manufacturing
industry demands “First Part Correct”.
In the case of High Speed Machining (HSM),
the fast feeding movements can break the expensive
cutter easily. Traditional NC simulation is not good
enough because it only checks geometrical errors
and most of them checks cutting process graphically
without an in process model. Hence, they have
built-in difficulties to calculate the exact cutting
load. We believe that the dynamic machining load
has to be taken into consideration as it greatly
affects cutter life, geometry accuracy and the
surface finish. Further, the huge tool path data files
for HSM give challenges as well where millions of
lines of NC code are commonly encountered. The
traditional NC verification is so slow that even
HSM itself is faster than verification. The size of the
program combined with a high feed rate makes it
almost impossible to run test simulations prior to
cutting.
In recent years, Singapore Institute of
Manufacturing Technology (SIMTech) has
developed a more efficient approach based on a
geometry representation which has been patented.
The system starts with a solid model of the

2. machined part, and can simulate and optimize
machining processes quickly. NC codes could be
reversely post-processed into graphical 3-D tool
paths, and interactively viewed, edited, and
optimized. The user can highlight or hide
operations, tool paths, or layers. The user can also
edit the tool paths in a certain layer. Tool paths and
cutting results can be viewed from any viewpoint
and checked graphically. The machined part and the
design part are compared for the remaining stock
and the over-cut areas. Error-free optimized tool
paths can be achieved. Hence this application
eliminates the need for a time-consuming test cut. A
quick and simple post processor is used to translate
the error-free tool paths into NC codes. This
technology is suitable for machine tool NC tool path
simulation, verification and optimization in the
precision engineering, automotive, aerospace, and
electronics industries.
Based on this technology, SIMTech has
developed several practical applications for mould
manufacturers. These include QuickSeeNC (a
machining simulation and verification tool) and
PartingAdviser (a plastic part analysis tool, which is
to be introduced in a separate paper). Both of them
provide “What You See is What You Cut”
(WYSWYC) functionality for CADCAM engineers
and shop floor machine operators. They could be
integrated with other CAM software through a
native APT adapter.
2. NC PROGRAM ERRORS
The average scrape rate for local mould maker
is 15%. There are different kinds of NC program
errors. Some of them are caused by human errors
which depend on the NC programmer’s experience.
These errors can be classified as: (1) Inaccurate
results in the patching up of surfaces; (2)
Inappropriate selection of machining process; (3)
Mistakes with the references to the part face, the
checking face and the selection of boundaries; (4)
Incorrect machining set-up; (5) Unusable cutting
parameters. However, a large amount of errors are
caused by the systems or the algorithms deployed in
the shop floor machining.
In the current industry practice, different
CADCAM systems are used among business
partners. One of the main problems that cause low
quality in solid models in the current CADCAM
approach is the data translation process which is
required due to the poor interoperability among
different CADCAM platforms. The low quality of
solid models cause low quality NC programs with
following three reasons: (1) The post-processor
generates arcs with large radius (several meters),
which will cause errors in some CNC controllers;
(2) The surface deformation during to data exchange
problem; (3) The surface tolerance is too low in
surface model translation; for example, it may be
larger than 0.1mm. Figure 1 shows a set of typical
parallel tool paths which generates uneven water
level cutting. (4) The data defects cause unexpected
NC problems. For example very often, there are
gaps between surface patches after data translation.
Due to such patch gaps in the part surface geometry
model, sudden plunge could be generated as also
shown in Figure 1.
Fig. 1 Typical NC toolpath errors
In another aspect, the NC tool path generation
algorithms are unstable because that (1) the tool
path planning is a very difficulty task; (2) the tool
path sampling method varies considerably from one
CAM package to another; and (3) there are no exact
offset solutions for curves and curved surfaces.
However, comparatively, the most significant
problem for NC optimization is the lack of In-
Process Model (IPM), so the tool path generation

3. with the CAM application does not consider the real
cutting geometry. This causes a lot of process
problems, such as air cut and gouging. It can be
observed that with the tool paths directly generated
from CAM application, the cutter moves through
remaining stock too fast, or too many conservative
air cuts in semi-finishing and finishing cut, or near-
touch cutting conditions occur which cause fast
cutter wearing. The recent development of high
speed machining (HSM) requires huge tool path in
order to realize the constant load. The constant load
is a very important factor in HSM. The cutter will
break under uneven cutting force. As the feed rate is
already very high, it does not reflect the sensitive
cutting situation to verify the NC program with the
traditional test cutting. It is highly demanded to
check the dynamic cutting forces before the
operation. Hence, there are strong requirements
from industry for a real dynamic simulation. The
simulation results can be used to optimize
feed/speed. Graphics-based visual tool path
checking is difficult as the tool path overlapping
with each other. Besides, it takes a long time to run
traditional simulation software.
3. FUNCTIONAL REQUIREMENTS FOR
AN IDEAL NC SIMULATOR
Theoretically, NC Simulation should feature
the full 3-D solid model of the entire NC machine
tool with detailed simulation for the manufacturing
(e.g. material removal) process. There are four
aspects to be covered by a comprehensive NC
simulator:
(1) NC tool path visualization. The simulator
enables programmers and machinists alike to
preview exactly what will happen on the machine,
especially for the cutter movements, or even for the
shop floor. Many users apply NC simulation for
electronic shop floor planning and documentation.
(2) NC verification. This function detects
problems in the NC tool path program. It is a
powerful visual inspection tool, which highlights
fast feed errors, gouges, and potential
crashes/collisions. Programmers can detect and
correct problems before prove-cut. Ideally, with NC
Verification you can virtually eliminate NC program
mistakes, greatly reduce the time spent on prove-
outs, and make the move to "lights-out" machining.
(3) NC analysis. This simulation function
identifies the tool path records responsible for an
error. You can quickly verify the dimensional
accuracy of the entire part with a full array of 3-D
measurement tools. NC analysis function compares
the simulated in-process part model to the designed
part model and checks if the machined part matches
to the design. One of the sub-function for NC
analysis is to perform constant gouge checking.
(4) NC optimization. This function
automatically determines the best feed rate for each
segment of the tool path based on the machining
conditions and amount of material removed.
Optimizing NC feed rates can greatly reduce the
time it takes to machine the part, assure the cutting
safety, and improve the quality of surface finish.
4. Quick NC SIMULATION METHODS
Z-Map [1] geometrical representation is widely
used for NC simulation verification. A map of Z values
represents the object geometry. Z-Map representation
can be effectively used for the surface that is always
visible from above in the direction parallel to the Z
axis. Since 3-axis milling parts are composed of
surfaces visible in the z direction, they can be
effectively expressed by the Z-Map representation.
With the Z-Map representation, the machining process
can be simulated by intersecting the Z-Map vectors
with the cutter profile.
Since the precision of Z-Map is decided mainly
by the XY resolution along the vertical walls, how
to increase the resolution along these walls and
reduce memory becomes a critical issue. An
important feature of 3-axis milling came into play.
Viewing from the top, the vertical walls only cover a
relatively small percentage of Z direction projection,
can we use finer resolution along the vertical walls
while maintain a rough resolution in the planar
area? This was the initial motivation for our
extended Z-Map method [2].

4. The size of the grid can be reduced through
using sub-cells. In this study, at least one grid on a
Z-Map is segregated into sub-cells. Only grids
corresponding to intricate features on the surface of
an object are assigned sub-cells to improve
representation of object features. Figure 3(a)
illustrates the plan view of the z-map grid with sub-
cells 52 (a level finer than 44) while Figure 3(b)
hows the front sectional view.
(a) (b)
Fig. 3 Extend Z-Map sub-cells along vertical walls
However, the precision of XY dimension is still
limited by the size of sub-cells. For a sub-cell of
0.1mm, the best precision is 0.1mm in XY plane.
There is a need to represent XY dimension more
precisely. Instead of using vectors in the sub-cells,
we use sticks in the sub-cells that have volumes and
surface geometry (see Figure 4). A B-rep surface
model can be represented precisely using a map of
B-rep sticks.
Fig. 4 Stick geometric representation method
Milling simulation in stick method involves
Boolean subtraction operation of the cutter from
each stick. Figure 5 shows different stick shape after
cutting. The experiments with B-rep stick model are
very slow and create a huge B-rep model. To
simplify the stick representation and Boolean
operations, a wire frame polygon instead of a real
solid with surfaces is used in a stick cell. This could
greatly increase simulation speed. The data structure
of polygon is much simpler than that of B-rep which
needs a group of complicated pointers to maintain a
double wing data structure.
Fig. 5 Different shapes of stick elements
In the real world, part surface objects are not
always uniform in XY dimension but can be any
shape (see Figure 6(a)). Nodes can be flexibly used
to enhance sub-cell’s precisions in representation of
object face. For example, one edge of the sub-cell
may have two overlapping nodes to represent a
vertical face. The nodes of a sub-cell may not be
uniformly distributed over XY plane. Figure 6(b)
shows a close-up view of a portion of the Z-Map
grid with nodes labeled with 54 (level number).
Figure 6(c) shows how a circular hole is represented
with adjusted nodes. Figure 6(d) shows the stick
method that has more sub-cells over the vertical
walls.
(a) (b)
(c) (d)
Fig. 6 Extend sub-cells to approximate a vertical wall
Based on this patented geometry
representation, we developed a more efficient NC
simulation system. The system starts with a solid
model of the machined part and quickly simulates
and optimizes machining processes. NC code could
be reversed by a generic post process into 3D tool
44 52 5252

5. paths, which can be further graphically displayed
and interactively viewed, edited, and optimized. The
user can highlight or hide an operation, tool path, or
layer. The user can also display and edit a certain
layer of the tool paths. Tool paths and cutting results
can be viewed from any viewpoint and checked
automatically. The machined part and the designed
part are compared for calculating the remaining
stock and any over cut. Error-free tool paths are
created, eliminating the need for a time-consuming
test cut. Quick and simple post processors are
developed to export the optimized tool paths into
NC codes.
Figure 7 shows an application example of a
steel insert of ‘discman’ (a portable CD audio
player) mould from a company. There are 16
operations in this CLS file. The resulted NC codes
contains almost half million of NC blocks.
QuickSeeNC loaded in half million lines in seconds
and displayed tool paths in different colours
according to operations (see Figure 7-9).
NC Program Data Sheet core_insert_nc
No. Name Cutter SOL Time
1 V64C61CA D20.00 R0.000 L75 96m
2 V61CA1 D20.00 R0.000 L75 11m
3 V64C61CB D12.00 R6.000 L75 31m
4 V64C61CC D6.000 R3.000 L75 45m
5 V64C61CD D6.000 R3.000 L75 153m
6 V64C61CE D4.000 R2.000 L75 33m
7 V61CE1 D4.000 R2.000 L75 11m
8 V64C61CF D20.00 R0.000 L75 2m
9 V61CF1 D20.00 R0.000 L75 1m
10 V64C61CG D8.000 R0.000 L75 6m
11 V61CG1 D8.000 R0.000 L75 7m
12 V61CG2 D8.000 R0.000 L75 7m
13 V61CG3 D8.000 R0.000 L75 7m
14 V61CG4 D8.000 R0.000 L75 7m
15 V64C61CH D16.00R0.000 L75 13m
16 V61CH1 D16.00 R0.000 L75 14m
Total NC Program = 16
Machine time = 7 hours 23 minutes
Fig. 7 An example of 16 milling operations
Fig. 8 Toolpath of half million NC blocks
Fig. 9 Extended Z-Map simulation
Another patent [3] has been filed to use this
extended Z-Map representation for quick plastic
mould design and rapid tooling application. A CAD-
independent mould design toolkit has been
developed based on this patent.
Fig. 10 PartingAdviser for quick concept mold design
5. TOWARDS PERVASIVE SIMULATION
Despite the fact that simulations are becoming

6. more widely used, there is a lack of integration of
simulations of manufacturing processes. Most work
focuses on an individual process being simulated.
Significant gaps remain in M&S technology –
particularly in the provision of a general in-process
model (IPM) that can be integrated across diverse
manufacturing processes [4]. To overcome the data
exchange problem between different models, we
extended the concept of the IPM from machining
process to other manufacturing process and intend to
develop a unified in-process geometrical model for
multiple machining and layered manufacturing
simulations. Figure 11 shows an application example
of a voxel model.
Fig. 11 Experiment result of voxel model
Further analyzing the voxel model, the authors of
this paper believe the voxel-based volume modeling is
a very promising approach to the unified IPM for
multiple machining and layered manufacturing (LM)
simulation [5-6]. As a natural clone of the LM
technology, the voxel model of an object and the object
fabricated using an LM closely resemble each other
since both are made of layers of small cells It
eliminates the STL format and eases accomplishments
of tasks such as estimation of errors in the physical
parameters of the fabricated objects, tolerance and
interference detection. Furthermore, voxel based
models permit the designer to analyze the LM object
and modify it at the voxel level leading to the design of
custom composites of arbitrary topology.
6. CONCLUSION
NC errors could destroy work pieces, even
damage machine tool. In the age of small batch
production, there is no time for trials and errors. The
challenges also come from huge tool path of HSM.
The traditional NC verification is so slow that even
HSM itself is faster than verification. The size of the
program combined with a high feed rate makes it
almost impossible to run test simulations prior to
cutting metal.
In this paper, we introduced our recent research
results on the quick NC simulation, which can
generate finished machining geometry efficiently
and accurately; at the same time, in-process cutting
volumes are calculated which allows for cutting
conditions optimization.
The novel hybrid multiple machining and
layered manufacturing processes posed a new
challenge to process planning and verification. It
seems that the voxel-based approach could be a
promising solution. More research work is to be
carried out. More detailed description will be
introduced in future papers.
Reference
1. R.B. Jerard, S.Z. Hussaini, R.L. Drysdale and
B. Schaudt, 1989, “Approximate methods for
simulation and verification of numerically
controlled machining programs”, Visual
Computer, 5(4), pp. 329–348.
2. P.L. Liu et al., 2002, An object representation
method, WO04032001A1
3. P.L. Liu et al., 2002, Mold design method and
system, WO04042481A1
4. www.imti21.org/msam/UnitProcess.pdf
5. Chandru, V., Manohar, S., Prakash, C. E.,
“Voxel-Based Modeling for Layered
Manufacturing”, IEEE Computer Graphics and
Applications (1995), v.15 n.6
6. Jang Donggo, Kim Kwangsoo, and Jung
Jungmin ,”Voxel-based Virtual Multi-axis
Machining”, International Journal of Advanced
Manufacturing Technology 16(10), 709-713,
2000

7. more widely used, there is a lack of integration of
simulations of manufacturing processes. Most work
focuses on an individual process being simulated.
Significant gaps remain in M&S technology –
particularly in the provision of a general in-process
model (IPM) that can be integrated across diverse
manufacturing processes [4]. To overcome the data
exchange problem between different models, we
extended the concept of the IPM from machining
process to other manufacturing process and intend to
develop a unified in-process geometrical model for
multiple machining and layered manufacturing
simulations. Figure 11 shows an application example
of a voxel model.
Fig. 11 Experiment result of voxel model
Further analyzing the voxel model, the authors of
this paper believe the voxel-based volume modeling is
a very promising approach to the unified IPM for
multiple machining and layered manufacturing (LM)
simulation [5-6]. As a natural clone of the LM
technology, the voxel model of an object and the object
fabricated using an LM closely resemble each other
since both are made of layers of small cells It
eliminates the STL format and eases accomplishments
of tasks such as estimation of errors in the physical
parameters of the fabricated objects, tolerance and
interference detection. Furthermore, voxel based
models permit the designer to analyze the LM object
and modify it at the voxel level leading to the design of
custom composites of arbitrary topology.
6. CONCLUSION
NC errors could destroy work pieces, even
damage machine tool. In the age of small batch
production, there is no time for trials and errors. The
challenges also come from huge tool path of HSM.
The traditional NC verification is so slow that even
HSM itself is faster than verification. The size of the
program combined with a high feed rate makes it
almost impossible to run test simulations prior to
cutting metal.
In this paper, we introduced our recent research
results on the quick NC simulation, which can
generate finished machining geometry efficiently
and accurately; at the same time, in-process cutting
volumes are calculated which allows for cutting
conditions optimization.
The novel hybrid multiple machining and
layered manufacturing processes posed a new
challenge to process planning and verification. It
seems that the voxel-based approach could be a
promising solution. More research work is to be
carried out. More detailed description will be
introduced in future papers.
Reference
1. R.B. Jerard, S.Z. Hussaini, R.L. Drysdale and
B. Schaudt, 1989, “Approximate methods for
simulation and verification of numerically
controlled machining programs”, Visual
Computer, 5(4), pp. 329–348.
2. P.L. Liu et al., 2002, An object representation
method, WO04032001A1
3. P.L. Liu et al., 2002, Mold design method and
system, WO04042481A1
4. www.imti21.org/msam/UnitProcess.pdf
5. Chandru, V., Manohar, S., Prakash, C. E.,
“Voxel-Based Modeling for Layered
Manufacturing”, IEEE Computer Graphics and
Applications (1995), v.15 n.6
6. Jang Donggo, Kim Kwangsoo, and Jung
Jungmin ,”Voxel-based Virtual Multi-axis
Machining”, International Journal of Advanced
Manufacturing Technology 16(10), 709-713,
2000